Full auto-generated transcript below. Beware of typos & mistranslations!
Daniel Brady 3:32
helping me out, it's just like, gotta make another post, you know, I just, it's so unnatural to me, I can find in person and finally giving talks and like, Yeah, I kind of stopped, but just like, putting yourself out there, you know, with like, a little bit of a daily feedback is very new to him.
Nick Jikomes 3:45
Yeah. I mean, for me, when I started it, I just decided, like, Okay, I'm gonna do this, if I'm going to do it, like, I'm going to do it. So I committed to like, one per week. So like, when it first started, you know, I had no audience or anything. And so, you know, I had to really try to get people, but actually not that hard. Like most people, like, I target mostly scientists who are really happy and enthusiastic, like, Oh, someone wants to hear about my work. Sure. But even you can really punch above your weight easily. Like if someone has a book or a show coming out. Right? As long as they don't think you're like a psycho or something, write them a reasonable email, like they're probably gonna say yes. Right. Yeah.
Daniel Brady 4:25
And I also think too, like, given your background, and just like just having a science background to I think it makes it more approachable. You know, we're talking about potentially complicated topics, both for general audience and, and I think that sort of expertise is just kind of really desired, especially with all this like general AI hype and stuff like that, like, yeah, you know, I went to like this talk yesterday, where people were, they were basically like, you should put yourself forward like as someone who does machine learning, who has a PhD in neuroscience, like people want to hear from you about this stuff. I'm like, but there's so many people just like talking about it all the time. It's so annoying, you know? Yeah.
Nick Jikomes 5:00
Yeah, no, I mean, but I completely agree with that. Like, especially with your background, and you know, just the ability to articulate things. There's so much noise out there that people find signal. I think they really they really attach themselves to it. Yeah.
Daniel Brady 5:14
Yeah. I think that's honestly, like one of the biggest things about gender bias is how much we're gonna need to find signal again, basically a bunch of noise.
Nick Jikomes 5:22
Yeah. And I just like, I also like, I just started, like, recording everything I reasonably could. So like, if someone invites me to give a talk, like online, I just say like, Cool. I'll give a talk. But I want I want the recording.
Daniel Brady 5:33
Okay. Yeah. Is that something that we can also have one from this talk?
Nick Jikomes 5:37
Oh, yeah. Yeah. Okay. You can take whatever you want from it. Okay, awesome. Um, do you want to just tell everyone a little bit about your science background and what your training was?
Daniel Brady 5:47
Sure. Yes. So Well, my name is Daniel Brady. I usually go by dB, or Dan, and so like, are venerated hosts, Nick, I have a background in neuroscience. I used to be a neuroscientist for a while. I did my undergraduate at Berkeley, where I studied psychology and neurobiology. And then I got my PhD in neurobiology from Harvard. And after a brief stint of a postdoc at UCSF, I decided to move over into the data science and machine learning world in tech. And so I've been doing that now for about the last decade, mostly in the E commerce space. Now I focused on direct to consumer companies. But yeah, it's all about bringing science to business.
Nick Jikomes 6:32
Interesting. What did you study for your thesis? Tell people about that, because it was one of my sort of favorite projects that I saw when I was there. Yeah, so
Daniel Brady 6:41
you know, I, I studied. So my lab I worked in into Cow hunches lab at Children's Hospital, Boston, and the lab more generally studies how early life experience fundamentally shapes the way in which your brain processes information for the rest of your life. And my project, which was kind of a new direction for the lab was what happens when you lose the sensory system early in life. And what happened to the parts of the brain that were supposed to process that sensory system, while they start to get taken over by the other ones. So we studied in a mouse model, if mice didn't ever have vision, right, they never use their eyes, then the visual part of their cortex would start to respond to sound, and presumably touch but we just studied sound. And we basically studied the molecular mechanisms, and anatomically, what was going on in order for what was supposed to be your visual cortex process sound? And how that like how that gets shaped? That's roughly what we
Nick Jikomes 7:34
get. So is that why like, you know, you always hear things like, people who are born deaf, their brain, you know, the visual parts of the brain respond to the auditory inputs and things like that. So that's true.
Daniel Brady 7:47
Yes, that is true. And it really matters upon like, exactly when you lost that sensory system early in life. So So you know, if you lose it, when your teenager or as an adult, then you'll have basically, that won't happen at all. But if you lose it very early than then you do a lot of your brain, what your brain basically does is it first has this kind of genetic mechanism to pattern roughly what it's supposed to be laid out as. And then, and it happens in a hierarchical hierarchical way where different regions, your brains at different times, starting from the lowest level processing to frontal stuff, basically say, Okay, I've been like set, I'm now ready for the environment to actually shape how I process information, right? So you have large, our brains have a large part of it dedicated to vision, right? And it's a lot of unused space, if you don't see, because if there's something's wrong with your eyes, so then there's a period of time where your brain goes, Okay, I set all this stuff up for vision, am I actually use any eyes? What are the statistics of the visual world that I'm experiencing, I'm gonna sculpt all those connections in order to better process that information. And that's kind of what we studied. And what's really interesting is that this, this has been linked to why people who lose sensory systems early in life tend to be better with the other ones, right? So someone who loses their vision can usually pick up sounds at lower frequencies or like at faster, you know, iterations of whatever, and that gives them a little bit of a kind of Daredevil ish sort of superpower. And it's because they have a lot more of their brain area dedicated to processing that sensory system. Yeah, I
Nick Jikomes 9:18
think it's pretty intuitive to people that, you know, the earlier something happens in development, whether it's something going wrong, or some kind of experience that you have, it has a much larger impact than if the same thing happens later in life. But But why? Why is that exactly? Why is it that early development, you're you have sort of more plasticity and more ability to be influenced by the environment? And then eventually that kind of goes away?
Daniel Brady 9:42
Yeah, so there's a couple Well, molecular anatomical reasons for that. But basically, as I said, like there's this kind of gendered patterning where you have like, all these neurons that are connected to each other in a certain way. And then there's this kind of molecular switch that occurs that allows your brain to be plastic Right. And then what happens is that it starts reorganizing itself. And then what occurs later is what's called myelination. And we can just think of it as kind of like laminating the neurons in place. And so they don't have they physically don't have the ability to move and connect as much as they did when you're younger. And that kind of like that, plus some other molecular mechanisms allow it to kind of sort of freeze in place. And then in order to then really have significant changes, after that, you have to really engage in what are called your neuromodulatory systems. And so these are like the top down processes, right. So for children, they can kind of actively, they can just kind of passively experience the world. And their brain will start being shaped by that. But for me to like learn language, like I have to study for a really, really long time. My daughter was like, pick up new words like every day, right? Without Really Trying.
Nick Jikomes 10:49
Yeah, interesting. And like, so what are some examples of like these critical periods? I know that you guys studied the visual system. But like, what, what are some examples that we can give people to just help them think about this?
Daniel Brady 11:02
Yeah. So one classic example, which most people in my former lab study was about the problem called amblyopia. So this is when someone is born, typically with a cataract over one or one of their eyes. And so the actual eye itself is fully functional, but it's occluded. And so you have, you have space in your visual cortex and your primary visual cortex that responds preferentially to one versus the other. And then that kind of experience is like, Okay, this is a column of visual neurons that respond to the left eye, this is a column visual neurons that respond to the right eye. When you have that occlusion early in life, because you're not getting input from one of those eyes, what happens is that the other eye starts to take over. And so if you do if that happens early enough, and let's say you don't treat amblyopia until you're 10 years old, or something like that, which is common, especially in the developing world, you can remove the cataract. But your central nervous system, your brain has already been like, I've already dedicated all that space to the eye that was functioning at that time. And so you have a permanent cortical loss of vision, even though the eye itself was functional the whole time, it was just being covered. And this is also true of people who, whose eyes are not perfectly aligned, who have like lazy eyes and like that. So that's why you oftentimes see corrective measures early in life, or they put a patch on the eyes to kind of like get them to orient correctly again, otherwise, once again, you can lose vision in one of your eyes, even though it technically works just fine.
Nick Jikomes 12:32
Yeah, so things really do get locked in at some point in development. That's right. And is that like, Is that similar to why we learn like, like you mentioned, your daughter, y, like little kids seem to sort of best effortlessly and passively pick up language. But if you try to learn when later, or you didn't fully learn one, when you were young, you never like, get full, full fluency.
Daniel Brady 12:54
Yeah, that's correct. I mean, there's there's lots of different reasons that language is very complicated for for why fluency is not really that easily achievable. When you try to learn language later in life, some of it is basic as processing the sounds themselves. So there's phonemes are like the basic units that make up language, like a plus sound of book, all that kind of stuff. And so we actually have the ability to hear all the different phonemes possible when we're just born. But in any given language, not all of them are used. And so during the first year of life, what happens is that a baby's brain as its learning language, and trying to figure out what the hell's going on, it starts to say, like, hey, you know what, this, our sound and this L sound, don't really make a difference. So I'm gonna like perceptually, collapse that space. And then later in life, if you try and learn another language, so for example, arnelle and say, Japanese are not really distinguishable, then you have essentially, what is an accent? Because to you, you can't hear the difference between rumor and liver. Because there's a baby, your brain was like, No, those are not different sounds in the context of the meaning of what I learned it. And so that's, that's an example of just on the basis of hearing it. There's also that first beginning as well.
Nick Jikomes 14:04
I see. So when someone has an accent, it's basically because in early life, there was no need in their native language to distinguish between two sounds like an R and L. And that gets sort of locked in because there's no there's no information there for them in that context. And later in life, they try to learn that but their brain is already sort of in incapable literally of distinguishing those sounds.
Daniel Brady 14:26
That's correct. And what's actually really interesting about this and Patricia Coles, one of the main researchers who studies this is that like, that collapsing, which, you know, at first glance, sounds like something that is like why would you want to do that? It actually helps you learn those languages faster, right? Because if your little baby brain is hearing River, and then I pronounce it river and then liver, but I'm mean the exact same thing. And it's gonna be very confusing because you're going to hear all the slight variations when the adults speak. And you're going to hear all these differences in the way But you're just supposed to say like, oh, this is actually the same word, right? And so you're gonna clap. So it's actually easier for you to learn a language. If you start figuring out which phonemes don't really matter, in a sense, and we have plenty of ones, ourselves as English speakers that we don't use, you know, like in India, there's a lot of what are called retroflex retroflex of phonemes. So, if you say, duh, and then if you say, duh, right, and that actually sounds to us like an Indian accent, but it's actually really hard for us to hear. So if I say dairy or dairy, that probably sounds very similar to people who don't speak one language in in India. But that is a difference that they happen. So lots and lots of different accents are because of this kind of perceptual collapse of different phonemes.
Nick Jikomes 15:43
Interesting, what if someone, what if someone grows up bilingual and they learn two languages from the start, maybe one parent speaks one and the other one speaks a second one, and they hear both, they become fluent in both does something happened developmentally, where they can then pick up languages as an adult even faster?
Daniel Brady 16:01
Yeah, that's, that's, it's a very interesting concept. And it's, it's something that is like kind of loosely studied, there is an idea that that might be the case. So what definitely can happen is that if you have language, especially if they're very different languages, or one of them's like a tonal language, like Mandarin, or something like that, you still have a lot more of a space available, right. So if you speak something like Russian, which has used a lot of different types of phonemes, and then you speak a tonal language, like Mandarin, or something like that, you kind of have a lot of space, to hear a lot of different sounds that like if you speak say, like Hawaiian, or something else, where they have actually a very few subset of the actual phonemes, it'd be very, very difficult for you to pick it up, because that stuff is all gone. And it's probably it's almost certainly the case that like, knowing more languages makes it easier to learn languages. Also, just because you've, you've seen, you know, what are the tips and tricks, essentially, of learning different languages and like how to get by when you do that. So it's definitely very important to kind of learn languages, one of the major things that, you know, for people in our, in that space, who stayed this sort of stuff is like, we're all very strongly believe in early childhood education, right. And then we also have this kind of like, weird, double edged sword, where it's like, early like expect is super important. But then it's also like, we're also super resilient, right? Babies are really resilient. Children are really resilient. And, you know, the key is just to like, make sure that they have a good happy home, have good educational experiences, and stuff like that when they're when they're young, but they can really, they're very resilient, especially on the different sorts of things that they can experience as long as you do it early enough.
Nick Jikomes 17:29
Yeah, the other thing I think about here, too, is, you know, when you talk about like, an animal, a mouse or a human, maybe they're born blind, or something, that visual part of the brain gets taken over by some of the other sensory systems. So they're dedicating more real estate to listening, because they rely on that more say, What about something like synesthesia, does that tie into this at all, where like, some of these sensory systems kind of get crosstalk, and people, you know, hear things that they feel physically and things like that?
Daniel Brady 18:01
Yeah, it's one of the more interesting concepts of what our work was. So So I work with this other graduate student named Liam. And so I did a lot of what's called the Systems Neuroscience. So I was actually recording from neurons and how they process information. And she took a more of anatomical approach. So one of the basic questions that we wanted to ask was, so how does this part of the visual cortex become multi sensory? Are there connections there initially? Or are there connections are Is it because like, the auditory cortex is close to the visual cortex that kind of gets invaded by its neighbors. And what was very interesting from our findings is that we actually found the found like an A connection directly from auditory Thalamus, which is like a brainstem, to the visual cortex that is just there, weekly, but it's there, and it exists. And that normal visual experience prunes that, right. And so it's what's really interesting about this concept is that it suggests that babies brains are actually multimodal to begin with, and that there's a separation of your, of your senses, as you experience them in the real world. And if you don't experience the sensory system, then because they're multi sensory to begin with, then the other senses sort of take over. This is really relevant in the context of synesthesia, which is basically the fusing of different senses. So there's thought to be basically two types of synesthesia, a bottom up mechanism and a top down mechanism. So a bottom up mechanism is very low level processing. Right? Like, and it's an example was like, I see letters that are associated with different colors. That's called grapheme synesthesia. And so in that context, what we think is going on there is that like, there is, you know, direct connections between, say, an auditory part of your brain and your visual part of your brain, right, and what our work you know, basically suggested was like, this is there from the beginning, and that for a lot of people, paranormal experience, these things are kind of pruned and shaped. But then for some people, they get stuck, and they have these various forms of synesthesia. There's another type of synesthesia, which is the top down Then one, which is kind of more like, you know, this sound gives me the sensation of red, it makes me think of red, it's like a red red colored glasses as opposed to like directly seeing red. And that is kind of like a longer longer form synesthesia, which is basically something that like, goes all the way up to like the frontal part of your brain and then goes back down. And so that's also something that's also really interesting, too. And the whole reason why this connectivity and activity dependent finding is super interesting is because we study it in turn, we studied it in terms of very low level processing of sensory systems. But it's thought that like the same sort of critical periods aren't important. We talked about language, but it also can be important for social development. And that, if you have, you know, this imbalance of the activity in your brain, or the connectivity of the brain, that could lead to what we think of as neuro divergence, or neurodevelopmental diseases, such as autism, or schizophrenia, and stuff like that. And so those are issues, typically, with connectivity in higher reaches at least the symptoms that we care about, you know, and want to treat, or to help, you know, adjust our society to deal with those deal, those people won't make them have a better life. But the basic idea of what to have happens all over your brain.
Nick Jikomes 21:11
So, you know, when we think about, like, early development, the brain is super plastic, you know, babies are learning in ways their brains can change in ways that the adult brain can't. And there's this notion of critical periods and different systems at different times where you've kind of got this window of plasticity, and then it shuts off and change can still happen to some extent after that, but not like it couldn't before. Are there any ways to sort of reopen or reinvigorate that plasticity either with drugs or by behavior?
Daniel Brady 21:40
Yeah, that's, I mean, this is like the million dollar question, right. And so we can manipulate them in certain mouse models with usually what we do is administer different types of drugs, that kind of, kind of, so a lot of a lot of the more technical details of of what our lab worked on was that basically, there's kind of this imbalance of excitation of excitatory neurons and inhibitory neurons in your brain. And that's kind of what triggers triggers these critical periods to begin, right. And so if you can manipulate that balance, and you can kind of reopen plasticity in sort of different types of ways and, and you know, a good friend of mine whose lab I've written a bunch of code for, hear from him recently reached at Mount Sinai, his lab really studies about the closure of critical periods and kind of opening it and then also setting it in terms of social behaviors. But right now, you can essentially have mouse models that are plastic for their whole life. So we talked about a little bit about myelination and stuff like that. So some genes that are really important in myelination, if you get rid of them in a mouse, the mouse is able to be plastic for its entire life, which is very interesting. And there's also different ways that you can like administer drugs, usually, at the time, when I was doing it, he would literally have like little mini pumps that you would attach to their head and they their brain would be like infused with different types of drugs that affected those sorts of systems. And that would get it in terms of actual just like behavior and stuff like that. Right now we know that attention and these neuromodulatory systems like your cholinergic system, right? They kind of work from these more frontal or basal areas that project back to the rest of your brain, those usually engaged like attention, right? So when we attend to something, and we, we've worked really hard at learning language, it kind of stimulates plasticity in that way. And there's ideas that like if you stimulated those parts of the brain while you were learning, or if you use drugs to kind of enhance those sorts of systems that you should do this. And so that's kind of what a lot of the targeting is, I think a lot of work is right now working on, on how can we make it so that this isn't something that's super invasive, right? Because a lot of the stuff right now is like you need to do brain surgery, in order to like see me like these things are in in order to wash sort of chemicals. But is there you know, like a pill that you could take? Or is there some sort of training regimen or something else that like some sort of stimulation that you can do on the surface of your head that kind of encourages that sort of adult plasticity that we see but makes it stronger and longer lasting?
Nick Jikomes 24:04
And when you're studying the opening and the closing of these critical periods in the brain? And you use drugs to do that, what what kind of drugs? Are they typically that have that effect? What part what receptors and what effects to having at the molecular level? Yeah, so
Daniel Brady 24:19
you know, what we study in terms of, of what triggers critical periods, it's usually as I said, it's this balance of excitation and inhibition. So a lot of the drugs that are important for that are some of the benzodiazepines, so things like Valium and stuff like that. And so one of the fundamental findings that my PhD advisor and his wife was also a neuroscientist found was that like, you know, benzodiazepines or valium can trigger early onset of sensitive or critical periods. Right. And so, obviously, the sorts of drugs that pregnant women should take is very, very important. But he basically showed that like, you know, this is really something that you don't want to take when you're pregnant, because what's going to happen is that whole thing of, you're setting this window of plasticity up for the real environment to then shape what it's like, well, if you do that in utero what it should have been when you're born and seen, right, you might have like these detrimental effects to the way in which you process all sorts of things. So that's one of them. And, and then on the other end, there's the cholinergic. As I mentioned, I keep saying cholinergic modulation, there are drugs that affect those sorts of things. I mean, the most famous of them is nicotine, right? So there are nicotinic acetylcholine receptors. And those are very important in this sort of process as well. I do not condone taking nicotine, gum or smoking. But my smoker ex smoker, friends always love that and always tell her the friends like oh, I just need to smoke more when I'm learning and that somehow will do that I do not recommend that sort of stuff. I think there's a lot of downsides to doing that. But it is those sorts of drugs are really important. And also learning.
Nick Jikomes 25:49
So these critical periods are, in large part governed by this balance of excitation and inhibition. And depending on where that balance is set, you're either in this kind of critical period where plasticity is high. And then when that excitation inhibition balance changes, that plasticity goes away, or at least goes down. And so is that why things like benzodiazepines have an effect here, because they are affecting inhibition directly.
Daniel Brady 26:14
That's correct. That's exactly what they're doing. And so that's kind of what tips that balance for them to start. And then as I said, then, then what also affects how they close in a normal development brain is really this myelination. So you have you have these neurons that are have these axons, and they're connected to each other. And then you have what are called oligodendrocytes, which are basically as I said, they're kind of like the rubber that like insulates the wires of the brain. And so they also develop over time, too. And that is a whole different sort of process, right. And when they kind of myelinated along the, the along the axons that can also limit the plasticity, right. So if you have like, literally all this rubber, on top of like most of the axon, like, it's not going to be able to be like, Oh, let me go over there, it's gonna be like kind of stuck, right. And so that is also a really, really important process. And so some of the first molecular mechanisms that were discovered to basically keep plasticity open, we're actually on that process of like myelination in the brain.
Nick Jikomes 27:11
And you have kids, right?
Daniel Brady 27:15
I do, I have a daughter who's 19 months old.
Nick Jikomes 27:17
And so you know, based on the work that you've done, and what you studied, and neuroscience, all of this early development, stuff, that that influence at all, like how how you watch your daughter develop, or how you think about, like, what to expose her to in terms of experiences.
Daniel Brady 27:32
Absolutely, and I think, and it's, it's been an interesting, it's been an interesting role, because I'm also then Now diving into reading studies about, you know, from economists and stuff like that large scale studies on babies and things like that, and not knowing what I know, now, like, knowing from like a basic neuroscience perspective, it's kind of fun, you know, we definitely are very keen on exposing her to as many different types of people as many different types of interactions, different languages as possible, and really just kind of being very, very engaged with her, as opposed to like, having her sit and watch TV all the time, or, you know, all this kind of stuff, because those things are really, really critical. But then at the same time, because of my background, also don't flip out on a lot of the stuff that a lot of parents, you know, tend to worry about, because there's a lot of stuff that you can't really control, or that kind of even out over time, right to like, doing something, you know, all this sort of developmental steps that people think about, does your son or daughter recognize themselves in the mirror? When did they start speaking their first words, all that kind of stuff, there's so much variability. And when you study this sort of stuff, from a very basic perspective, you don't worry that like, you know, your daughter's two weeks behind on taking the first steps, but then he's really good at rolling around and really bad at you know, like, all that kind of stuff. You're just like, Oh, they're fine, you know, you have a much wider window of like, when you actually start worrying about those sorts of things.
Nick Jikomes 28:54
Is there anything that you don't do or that you, you don't expose her to that maybe a lot of people naturally do as parents?
Daniel Brady 29:01
Hmm, that's a great question. You know, I think I mean, we certainly, we certainly don't expose her to any sort of drugs or pass and things like that, right. So like, both my wife and I don't smoke, so she doesn't grow up in a smoking household. Like there's, there's a lot of things like that, like, basically a lot of chemical sorts of insults, that we kind of take care of things like, you know, what, when I was in grad school, what a lot of people on and also study we're like, how do hormones influence classes and stuff like that? So like, are is your are your children being passively, you know, exposed to hormones, in milk or in water? All those sorts of things are heavy metals, like those are the sorts of things that we're tend to be a little bit more careful on to like those because we both have biology backgrounds, and like those sorts of chemical things. And then, and yeah, those are probably the main the main ones that we kind of like, actually spend time really, really kind of considering
Nick Jikomes 29:59
you And so you're in New York City now? I am. Yes, I'm in Brooklyn. And okay, I was there a few weeks ago. What? So how did you get into what you're doing now? How did you make that transition from being a neuroscientist to what you're doing now? And what exactly are you doing?
Daniel Brady 30:15
Yeah. So we can talk a little bit about my journey into what I do, because it's kind of a little bit of a windy road. So I graduated, I defended my thesis in 2011. But left over in 2012, as I started a little bit of postdoc. And it was at a time when, when data science was a thing, but not super common, right. So some of the big companies like Google and Facebook had data scientists. And they're basically what they were doing is just scooping up people like me at the time, right? There wasn't a ton of like, master's degrees in data science, or poorly knew what it was, it was like, what we're going to do is we're going to like, take a bunch of PhDs who are sick and tired of in academia, and we're just going to let them loose on things that have statistics and accompany and like, let's see, let's see what they can do. Yeah. So I kind of fell into to Tech, I knew that it was I did a lot of programming and sort of computational work, both in my PhD and my postdoc. So I had that background, you know, typically when, like, from earlier generations of scientists, when you decided that you didn't want to be an academic anymore, the first thing that you did, especially you went to Harvard is you became like a McKinsey consultant. Right. And that's still it's like, oftentimes the case, but because of data science, and because of machine learning, there's like this option, especially if you knew how to program at the time, that you could also just kind of go into tech. So that's what I did. And I fell into working in startups, because I had some friends who are working at startups. And I was kind of doing some lab work writing some code for Hirofumi at Mount Sinai. And this one lab, this one company was just getting started. And this is around 2013 2014. And that was what I called, like the same day delivery wars, there's, we can go into what that was, but like, basically, there's a bunch of companies that were all trying to deliver packages last mile delivery all the same time. So DoorDash, Postmates, all those sorts of things, they actually started at that time, and they had a company along that that was in that sort of space. And I didn't know anything about that world. So I would just go out there and hang out and write code with them. And they asked me to do a project for them. And they had all this customer data of like routing, different packages, and all this kind of stuff like that. And I made like, I don't even remember what it was, but like a dead simple, little probabilistic model of like when they should have more careers or less careers on and they were like, Oh, this is really, really useful. And I was like, oh, yeah, that took me like an hour is really rough. I didn't like really try very hard. And right. But this is really useful. And so when they actually raised around, I was their first employee. And so And what was really fun about being in that world was that, like, I could use all the skills that I had as a scientist, which at the end of the day was, was being able to make models of the world be able to think probabilistically, scientifically, mathematically, but apply it into this business, into the business context, in this case, in terms of serving people different goods as we deliver stuff. So it was a lot of logistics. And that kind of gave me the bug. And the fun thing about working in the tech world is that it moves super quickly. So you know, within a few weeks of my arrival, because of some of the work I did our pricing model changed, we grew a lot all kinds of I don't want to be responsible for the growth by myself, there was a lot of other people who were helping with, and then you know, a year from that, we were completely out of business, you know, and they're just like, so fun compared to what I really liked doing my academic work. But you know, in that context, it was like, what fundamentally became my thesis, I discovered it, you know, midway through my third year, and then it was like, another three years proving that that was true. Whereas in like in tech, it was like, Oh, that's a good idea. Like, we're gonna release that to 1000 people tomorrow. And I was like, Oh, okay. And then you see everyone react to that. And people are like, this is the best thing ever. I hate this. And it's kind of addictive. So we kind of became addicted, that sort of thing. And so I've been in tech ever since. And I worked for another another startup after that one collapse. And then what I had been doing fairly recently was another data scientist. And I started a consulting agency where we helped young companies once again, mostly in E commerce, and direct to consumer brands, help become data driven, right. So that's, that's something that is kind of a word that a lot of people use. And there's a lot of whelming people who are very good at marketing, very good at designing products. And they collect all this information, right? Because we have all these systems on the internet to collect information, and they really want to make good data driven decisions. Are my Facebook ads working? Right? Is this product worth investing in? Are these customers valuable? Are they not valuable? All this sorts of stuff? Like how much time are we spending serving people? And there's a lot of mathematical questions that you could answer that look at it scientifically, and people really don't know where to get started. And so we started a company that helps brands do this, help them hire, hire people, all this kind of stuff did like all sorts of things, dynamic pricing, all sorts of different projects. But our goal was, if we did something over and over again, we would convert to C Corp, and make that as a standalone product. Basically, like, what do we keep running into now we can productize and then sell that on its own. And so we've we've had this, you know, we literally, we came out of our like kind of stealth mode. At the end of the year, we launched our self serve app, literally like a week and a half ago. So we just kind of like, we're here, right? We've had a couple, a couple of customers and pilot customers before that, but like we have now arrived. And the fundamental thing that we do is that we clean up and unify customer data, right? So the cheeky way to say it is that I say that we're like the garbageman of data, the sanitation engineers. So once again, companies collect an enormous amount of information, they can collect information on how you're interacting with their website, they're collecting information on how you're interacting with their customer service representatives, how you're paying for goods, all this sort of stuff. And what they're really trying to understand is like, did you like what they had? Are you coming back? How can they make their whole experience and a whole product better, but they use lots and lots of different systems do this, they generate lots of data. And there's no real way for them to connect it all together. And that problem of finding out who you are, as a customer, across all these different systems that have slightly different perspective of you, is actually one of the hardest problems in computer science. The more general terminology for this is called entity resolution. It was actually started by the Department of Labor and the Census Bureau, because this was a huge problem. When taking the census, you have all these census takers out there, they're collecting information, like, Hey, I talked to Nick and I was like, why talk to Nicholas? And it's like, they're addressing the same maybe there's two guys both named Nick that live there or not? Right? Like that's a problem. You know, it's, it's something that through our entire life experience we can get very good at, but to teach a program, how to differentiate, you know, Dan, Daniel, Dan Brady, right, Dan Brody is that misspell Brady, all that kind of stuff is really, really tough. And so. So you need a lot of math and a lot of data science and machine learning and engineering in order to kind of solve those problems. And to be frank, it's something that is usually well beyond the capability of even the largest companies that are out there. And so we thought it makes sense as a standalone product, like you submit your data, we'll do all the hard math in machine learning to clean up and to connect all that data together and be like, Hey, this is Nick. This is Daniel. Right? And then we give that back to you. And then for the companies that we use for they have a lot of different use cases for this. One of the most basic things is like how many customers do they have? Because one of the fundamental double edged swords of the internet is identity. Who are you, you can present yourself as many different ways. Obviously, there's huge implications politically. We know everything that happened with Facebook, and the elections and all that kind of stuff, right. And those are very well rounded, talked about ones. But there's also ones like, well, you know, this young brand is like 15% off for your first customer. If your first purchase. And it's like, well, I'm gonna just make a bunch of fake accounts and pretend to pretend to be the first customer all the time. And most companies don't care about the 15%. But what they really care is, did they just get 10 new customers, or one person that bought 10 times? Right? That makes them think of very, very different ways in which they're thinking about the traction of their product of of how much in how to market it, everything, how to talk about it, what's working? And then also just kind of like, is this something that people come back to, right? And so all of those basic ways in which they look at it is wrong if that underlying customer data is wrong in and of itself. And so we're like, don't worry about that. We'll fix all of it up and give it back to you. And so that's what we're laser focused on.
Nick Jikomes 39:08
I see. So, like, I actually work for my day job at an E commerce company. And so I can I can relate to this. And I can, I can assure everyone that all of this is a big problem. But it's you know, if you think about it, right, like for those listening, you know, if I work for some, so I work at a company called Leafly. We've got an Android app, we've got an Apple app, we've got leafly.com. So you get people that sign up through an email, we might send out, someone might download the Android app, someone might download the Apple app, someone might do two of those and then have two accounts, or they might you know, not give us all of the information that would be convenient in identifying them. Or, you know, I work at a cannabis companies, something we have to worry about is are you at least 21 Right? And obviously, you know, when it comes to age gates, people lie all the time, or they just fat finger something and it's not accurate. And so you know, we've got all of these data streams. Just coming from this app and that app and the web, and you know, from this email campaign, and you know, all this other stuff, and there's all sorts of duplicates, we're missing information. And what you're saying is the fundamental problem here is a entity resolution or identification, who are the unique individuals in this corpus of data? And then to there's a lot of problems related to that, that have to do with, you know, cleaning the data, which means, you know, filling in stuff that isn't there or deduplicating, all that stuff. Yeah.
Daniel Brady 40:29
And then also just getting rid of junk, right? So the very first thing that we do is we spend a lot of time cleaning and validating all that information. Is this a real address? Is this a real phone number? Right? All those sorts of basic things, right? 123 Fake Street, okay. Don't use that, right. Because, once again, you know, if a company is like, Okay, we're gonna send out a bunch of catalogs out for because we're a clothing company, well, you want to make sure that they're sending them to places that really exist, that they're not going to get a bunch of postage back, right, that's also true of text messages. It's also true of notifications of, of, of emails, and so that that envelope, there are companies that are dedicated just to that, and that's really where we start. And we have to do that in order to really then use the math to put things all together, right. And then as you said, because they have all these different data streams, then we can give it back to them. And the final layer that we've we've started recently adding, which is kind of interesting, and I'm really excited about is that we specifically focus on direct to consumer brands right now. So like examples are like the Allbirds, the shoe, or skins, you know, slips that like brands like that, right? All sorts of different sizes. This because we laser focus on that data issues and connectivity problems are not in a vacuum. There are reasons why they happen. Right? And because we are ecommerce experts, and we have access to the data, we can then tell the branch why they had it. So it's like, Hey, you have these two accounts. And these are two accounts that should be merged together. And it's because they're taken advantage of or promo, or these accounts have been broken up, because it's actually a drop shipper. And it represents 50 people, or reseller, right, or it's like this is a bunch of Google bots that are checking in your prices, or these are scam bots trying to prove your system. Like there's really fundamental sets of reasons data just doesn't randomly, like slip and it's like, oh, this has changed to something else, is people or bots are trying to kind of take advantage or accidentally doing sorts of things. And so that gives companies context, even if they're not really sophisticated in terms of tech wise to make a decision, right? So the way that I say that from a business context is like, different matches have different business values. If I accidentally make two accounts, that's like, okay, maybe that's fine, we maybe we need to work on our UX or user experience to make that a little bit better. But it's like, if I'm making 50 different accounts to take advantage of something to commit fraud or something like that, that you really care about, then you can put an immediate stop to that. And that that also, that sort of context is really, really helpful to work with.
Nick Jikomes 42:57
And so like more concretely, like what does this look like? So if you're a direct to consumer brand, and you share some kind of data with your company, which is called what is it called?
Daniel Brady 43:07
It's called Rita O R. Ita.
Nick Jikomes 43:10
So I'm a DTC company, I give you a bunch of our data. What what do I get back? What is it? Is it a spreadsheet isn't an interface? What does it look like?
Daniel Brady 43:20
Yeah, so we have we have a bit of an interface and, and it isn't in a dashboard, that kind of like, gives you some of the top information that Casey brands care about. So the first is a score is better data good or bad. And the reason we have that is because we run this on a continual practice. And so what happens is that brands will like, say, start a marketing campaign, it'll see a bunch of orders and stuff like that. And it'll be like, okay, that gave us a lot of revenue. But like, did that mess up our data? And so the very first thing you want to see is like, just a general sense, is it good or bad, right? And we can talk about how we determine that. And then, but in that dashboard, it says, like, Okay, so here's things that you care about how many customers you actually have, and what is like being reported in your so your E commerce platform, however, if you just like did not do all that matching and hard stuff. And we do that for a bunch of the different metrics that they care about, we then go into what we call the root cause stuff, which is like saying how many users they have, how many? What are the exact opposite of like brand proponents, people buying for each other people who are like really kind of advocating for your brand. And then at the bottom section is kind of like a bunch of different for most of our brands, they like it in spreadsheets, right? Which is a bunch of ways in which they can then unify their different data sources. So we're like, hey, we look at all your data sources. You have a bunch of customers in here who want to get marketing stuff from you. But we checked your marketing platform, and they're not there. So here's a spreadsheet that you can like load their lists. And then here's another list of people that you can delete, because they're all duplicates or something like that. Yeah. So you're not
Nick Jikomes 44:46
just so you're not just giving people like, one unified list of here's all your unique customers, but you're actually you're actually helping them figure out what the problems are. You're telling them why they have duplicates, or why someone's missing, then they can update their own tech so that That happens less moving forward.
Daniel Brady 45:01
Exactly, exactly. So for the most sophisticated brands that have insurance like that, they really cared about that unified list that really connects all this stuff together. It's like that Rosetta Stone that they'll have in their data warehouse, or it could just be an Excel spreadsheet that they use VLOOKUP on, right, it doesn't have to be super fancy. But because of the different reasons and different ways in which data can be missing or duplicated across systems, we try and create different ways that they can like download or upload or whatever what they needed to do to like delete, or to unify all that data. That way, they can have all their systems that can cross talk appropriately.
Nick Jikomes 45:36
And so like when you started this company, so you and your co founders started it, obviously, you have the tech background and the startup experience to understand the problem space and how to how to build the tech. You've got your technical background, which feeds into that. How did you guys actually get going, though? Did you completely bootstrapped this? Did you have to raise money? And if so, like, how did you, you know, as the Harvard PhD guy? Yeah. And coming from that world, how did you navigate the the start the start start part of startup?
Daniel Brady 46:06
Yeah, it's a great question. I mean, so we had a lot of experience in terms of, of talking to people from our consulting days, right, which all of that was like word of mouth, you didn't really advertise yourself. So we had some good relationships with a couple of companies that were like, Hey, can we test this out on your, on your data, right, give us a proof of concept. And we actually are still kind of, in terms of funding, we were actually still in the very early stages we do are doing friends and family. So we haven't properly raising any money from VCs, we're, we're debating about whether or not to do and we're most likely going to do that. Probably right now, we've talked to a few that are interested and, and we'll probably see over the next few months, because we're like reaching that critical mass where we need funding to hire people because like, you know, it takes a lot of work to get something like this off the ground. But up until then we basically have been talking to close friends or families or other business advisors and kind of like getting a little community of people to support us into believing our vision, and they're just putting in, you know, you know, businesses are expensive. So this will sound like, you know, 10s of 1000s of dollars, right, but, you know, proper financial in the million dollar range, we talk to VCs, but this is more in the 10s to hundreds of 1000s of dollars, mostly with with our own local network. And it's been an interesting process. You know, as you said, like, we are both technical founders, both my co founder and I are machine learning engineers and data scientists. And so it's been a real interesting shift to kind of go from being that sort of expert in doing this sort of thing into like, learning how to sell, learning how to market, learning how to raise like, all the kinds of extra VC stuff, I mean, extra founder stuff that is required has been like the major thing that I've had to work on, over the last year, we made this product. With ourselves in mind, we're like, what is the thing that we hate doing the most, every single time we start with a new company, it's like we get the data. And then we got to clean and connect it all together until we actually get to the fun part of like, doing something with it, of activating on it and reporting on it. Right. And so, you know, we thought naively that like this is a product for data science and and engineers, and data scientists and data engineers love our product. But they typically don't have the budget or the wherewithal to actually be able to use us. So it's kind of been a very interesting experience to me, like, how do we talk to people in marketing, or sales, or operations or finance about it, and we can't use a lot of the crazy words, you can't use a lot of the technical jargon, right? That to show them how important this is, and like talking about the business impact of what we do. And so that's kind of what we've been doing and working on is really a lot of our messaging that the tech is obviously getting better all the time. But a lot of our fundamental ideas and the algorithms that we use, like we came up with a while ago. And it's really just more about learning how to talk about ourselves to the different audiences that we now interact with, they like finally to make to actually sell.
Nick Jikomes 48:53
So ignoring for a moment, the technical side, the numbers and the programming, and all that stuff. Do you feel like your PhD training helped you in any way in terms of learning how to market and sell and tell a story and things like that?
Daniel Brady 49:08
Yeah, absolutely. I think the I think what it helped the most with was learning how to talk about something that is potentially very, very complicated, in a simple and straightforward way. Right. So my goal as a PhD student wasn't like, explicit, I try to do this, but I always encouraged myself was like, Oh, here's a person who's asking me what, what I'm working on. And maybe the last time they took a science class was like, their junior year in high school. And that was 50 years ago, like how are we going to talk about this in a way that relates to them in a way that they can understand that doesn't intimidate them? And that is very, very useful skill. When you talk about machine learning and all this kind of stuff like that, you know, in that sort of context where there's a lot of noise in the field and a lot of hype to kind of bring those sorts of problems problems down to make it really really relatable. You know, I I, one of my my best friends from grad school, you know, he always used to make Have fun of me and make fun of himself. But he's like, Oh, my, you know, my parents really love your PhD. Because we understand what's going on. I'm like, Well, you study like ion channels, and like all this fun chemical engineering stuff. And it's like, very, very hard to understand. Like, I'm like, you lose the sensory system early on, like, you understand that, right? Like, like, these are sort of basic things. And like, when I work on my own stuff, it's like trying to find, through my practice with my PhD of like, being able to talk about it to a more general audience.
Nick Jikomes 50:26
Ya know, I found myself many times, you know, talking to people who will ask me like, Well, how did you go from doing a neuroscience PhD to what you're doing now? Because, you know, on one level, it's like, it has nothing to do with my training, not doing neuroscience now. But you know, one thing I've often told people apart from, like, the technical side of your training, and how those skills generalize, is like when you're doing hardcore experimental science, basically, everything you're doing is in some way, shape or form, preparation for telling someone a story. Yes. And so you get really good. And you get a lot of practice that like, How do I tell a story about something that's really, really complex?
Daniel Brady 51:03
Yeah, absolutely. And then I think another thing, you know, that everyone might be interested to learn about when you think about the process of science is that like, when you see these papers, when you see something written The New York Times, and it's like, oh, the sample size is 50 miles or 1000, people, whatever, well, there was a time when it was three. And the machines and your microscope work, like half the time, and you didn't have to do the full analysis. But you had to convince everyone in your lab, you had to convince the other friends and colleagues just like that, that this is worth pursuing. That we have an interesting idea. And it's really, really an evolving process. And then you do that many times for years. And then it's like, at the end, it's like, okay, well, you know, we've worked on this, we've kind of explored the scientific space and stuff like that, now, we just really got to build the numbers to make sure that we like nail was so significant is that we think that we have and like really nail everything down. But for a large part of the journey, you're really exploring how to do this. And that's extraordinarily simple. Similar to starting a company, I think that like, you know, now, having a company that I started in this space, I actually think it's probably very similar to starting the lab, right? In that case, the you know, your customers, and your funding, and it's different, you know, you're publishing papers, that's your output. And, and are giving talks, and, you know, your primary customer is like the government that's happy that you publish a bunch of papers to give a bunch of talks, right, and that you're gonna get really great people. But it's kind of a very similar process, like managing people meeting expectations, and, and really kind of learning how to carve a name out for yourself and that sort of space.
Nick Jikomes 52:35
Yeah. So, you know, if I'm like talking to the head of marketing, at my startup, what's the what's the elevator pitch for a reader?
Daniel Brady 52:44
Yeah, so you know, so one of the hard things about us is, is that there's such a fundamental issue is that it's, there's a couple different primary use cases, right? So depending upon your size, it could just be as simple as you're spending too much on your marketing, because you have a bunch of duplicates in your marketing list, and we will find them get rid of them, it could just be as simple as that, right. And if you're not a very sophisticated brand, like that could be an appealing enough play, that that's the case, if you are a brand that is more into personalization, and all that kind of stuff, then we will talk about that, right, you'll be like, you know, you don't want to send just generic sort of email blasts, right now, what a lot of companies care about is our personal relationships with their customers, they want to give meaningful recommendations to them, or meaningful upsells, they want to write content that engages them, that makes them aligned with a morally, or just like the vision of the company as well, as well as products. And the best way to do that is to have, you know, a full view of your customer. And so you can't have a personal relationship with your customer, if you just like don't know what they've done with you and don't know how they've interacted with you. And so that's kind of what I would often say in marketing. And then for performance, marketers weight, which is a subset of marketers who care very, very strongly about numbers, it's like, we'll give you the real ones will give you the actual customer acquisition costs, the actual lifetime value of customers, the actual repeat purchase rate, actual number of customers you have, and those are the numbers and they use all sorts of different formulas. And there's other numbers that they care about. But like that's how they make their big decisions on where to spend our cash. And like, for a lot of brands, they're completely off because of this fundamental issue. Yeah,
Nick Jikomes 54:17
no, and that's, that's absolutely true. Like, sometimes you don't have the information at all, sometimes you have it. And you might even know that the numbers probably aren't completely accurate, but you don't even know how to fix them. And you can really affect big decisions if those numbers change even a little bit sometimes. Yeah, and,
Daniel Brady 54:33
and the thing too, that's what people really care about. So there's there's been this very interesting shift in commerce, you know, where at the very beginning, then by beginning I don't mean thinking of all time, I mean, like 1970s, or something of that height, where what really mattered was what was called a product centric company, which is really typified by Apple. Right. Here's this amazing MacBook Pro. It has all these features I'm going to talk about it's really sleek. And it almost appeals to you like, you know, you want to be the type of person who uses a MacBook Pro. And it's like full capacity. Right? Then the next wave of companies are what are called customer centric companies. And these are companies that basically optimize every single interaction because it converted more people into customers and organs of Amazon is the best example of this every pixel, you know, on the screen has been fought and died for to make, like more people engage with with their platform where people buy from them, right? That's all well and good. But we now into this new stage where people really care about that customer centric, but they call it about relationship centric. A really great example of these are brands like chewy, so like the big retailer for pet foods, like, you know, when you stop a subscription, because your pet dies, they write you a handwritten note, right? It's like that's a meaningful interaction that builds a lot of loyalty. Right? And how do they know to do that, because they know their customers, they have full views of what they're doing, they ask them, they talk to them, they collect information about them, and they use it to be like, hey, you know, this person is, is, you know, canceling the subscription because their dog died, like, let's take the time to write them something to make them feel special, and be like, Hey, we're here for you when you have your next dog or something like that, right. And so we are moving very quickly into of space where basically all young brands want to be like that. They all want to feel very personal. And and they mean in in meaningfully do it. And then a lot of large brands are like hey, this is really important to us. And Nike does all sorts of things Disney to this is like, they know who you are. And you come to the stuff that they invite you to special things. They're like tailor made for people like you. And this is all through like decisions that they make. And so this this is like a kind of a core concept for where, you know, ecommerce and commerce in general is moving to.
Nick Jikomes 56:44
And yeah, I mean, so this is really exciting. What, how has it been like running the company? Like, are you doing this full time? Is this your full time pursuit right now?
Daniel Brady 56:54
Yeah, it is. It is. And it's super exciting. I mean, I think I think it's, it's a really, it's a really pivotal time to kind of be in this space, that the challenges of working, where I'm at is that it's a bit of what we call a category creation. So it's, it's a type of company that people in our space don't necessarily know that exists alone. So when we first started, we thought we'd have to spend a lot of time telling people that this is a problem. That's not the case, everyone knows that their data is really messed up. What they didn't know is like, Oh, I could tell this right now, in a very easy manner, assigning to a bunch of data sources and get good data back like that was what's missing. And so there's a little bit of education around there. But like, everything's post for that, it's been very easy to collect data, it's very easy to then activate on it. Well, we call data activate, which is like sending promotional campaigns, marketing, outreach, all that kind of stuff. So what's fundamentally missing right now, are very good ways to clean and process that information. So that when you collect all that information, and then when you act on it, that it's really, really good. And we are, we are really the first company in this space, right, that do this, there are companies that do this more broadly, right. But we are like, very specific of doing the wind y for clean data in in DTC, and that's kind of a new sort of thing. And so that's been very, very exciting to kind of do and to kind of be like, Hey, this is where we're hearing that people like, Oh, my God, that's been a big problem for mine forever, you know, so and so
Nick Jikomes 58:21
with the product with the service that a reader provides, and the output that you give to your customers? How long does this process take? How long does it take someone typically to get you the data that you need that they already possess? And then how long from there to do what you guys do? And then give them their insights?
Daniel Brady 58:36
Yeah, so the big question is, how much data are they giving us? Right? So that's the kind of thing but for most companies, it's, it's within a couple of days. Right? So people sign up for their data sources, we have a partnership with a company called Five Tran that handles all the authentication actually sending data over to us, right. So that's a large load off our plate and makes it really easy. It's a no code solution, which is really, really helpful for marketing teams. And that could take a few hours to a few days, depending on the data source of how much data they have. Right? So they have millions of transactions, it might take a day or two. And then for us actually do the cleaning, validation, and then the unification that also takes a couple business days. So we usually tell companies that like within a week, maybe two weeks, and we'll let you know, because you probably have a lot of data or it's really, really messy, or really funky things going on, you'll have cert clean data after that. And that's, you know, that's pretty fast. And any resolution is a really, really tough problem. And it takes a lot a lot of comparisons and a lot of computation. So being able to get an answer within like a week is pretty good. Yeah.
Nick Jikomes 59:40
So getting getting an answer within two weeks, maybe even a couple of days is what we're talking about.
Daniel Brady 59:44
Yeah, exactly. Yeah. If you're a new brand with I mean, we could do it in a few hours.
Nick Jikomes 59:49
Well, yeah, that's, that's actually really good. Yeah. So like, what, how much does this cost?
Daniel Brady 59:56
Yes. So right now because we can or more about just getting ourselves out there and being we're just like a flat pricing. So it's $250 a month. And
Nick Jikomes 1:00:09
then month by month, you can add more data to it and keep cleaning it. Yeah, exactly. Maybe Maybe from your product you can get, not just your your entity resolved customer data set, but you can figure out like, oh, we can actually make X y&z changes on our end. And now we don't have to worry about generating bad data anymore.
Daniel Brady 1:00:27
Exactly. That's the whole point. It's, that's why I use the analogy of kind of being dated garbage man or sanitation engineers is that like, it's not something that builds up, right, if you don't take out the trash, your house will become very dirty. But it's not something that ever really, truly goes away. Instead of technical debt, we refer to as data debt, right? You accumulate data that as you as you grow, and ironically, the faster and more successful that you are, the more data, bad data and messy data and siloed data that you generate, because you're trying all sorts of things. And so what we do is just a continual process, every month, we kind of clean up your data, and you now have another unified file that you can use, you can update it, and then you also have more and more context. And so if you make a bunch of fixes and changes, you'll see that reflected in our reporting, and then you can, and then you can then maybe try some things out. And so what some companies do is they say, like, Hey, we're gonna send you all this data, you're gonna clean it. And then we're going to do like our first type of this campaign. And like, we'll see, like, how did the data look like three months later? And like, evaluate what that what that looked like. And so by having this kind of benchmarking over time of your own data, you can kind of understand where you're at, understand how you're growing.
Nick Jikomes 1:01:35
And so like, now that you've started this company, you know, given your background, you've got a young child, you're in Brooklyn, New York. Is it scary at all that that you're doing this full time and starting a startup from scratch, given how fast things move? And all the uncertainty?
Daniel Brady 1:01:52
Yes, I mean, for sure, I think that, I think that it is scary. But it's more of like, it's more of just feeling like, this is all temporary. But how temporary is this is this is this because there's a lot of instability, there's a lot of wins when you sign a new customer, it's like a huge deal. And we hope to get to the point where it's like signing hundreds of customers a day or something like that, right? But and so I don't I have enough confidence, you have to have competence, or to be a CEO of company, I have no confidence that I feel like I think that this is going to be a very successful endeavor. And it's really just like, sticking through it. And, and knowing how long that kind of uncertainty is going to last. It definitely is finding but it's also super fun and super flexible. So you know, I work from home, and I can kind of set my own schedule, I work a lot, but then I also can be very flexible and hang out with my daughter. So in in many respects, it's a really, really great place to be.
Nick Jikomes 1:02:48
How do you? How do you prevent yourself from getting like burnt out or stressed and just sort of, you know, turning off work mode and things like that?
Daniel Brady 1:02:56
Yes. I mean, that's a great question, you know, so there's hanging out with my family. That's definitely the number one thing, right. So I have a daughter who barely speaks English, so she does not care about what I do. Right. She has no concept of it right. And so spending time with her is definitely fun and spending time with my wife as well. I'm also a big surfer. I live in Brooklyn, and have a surfboard in the background. So I tried to serve, you know, at least once a week, which you can do out here out by the JFK Airport in a place called Far Rockaway. And so that's kind of what allows me to keep saying, and so you know, when you live in New York, everything is expensive, everything is hard to do. So like your life kind of collapses into like, one or two hobbies and your family. And so like, that's kind of what really keeps me centered. And then yeah, and so that like, of course, it's like great restaurants, all that kind of stuff. So I make sure to really carve out that sort of time, so that I feel recharged. And I think the other thing, too, that you do, as a founder is that when you start you do everything, and then you start finding people, maybe they're working on a contract basis, and maybe you hire them full time, and they take and they're better than you at something, and you give them that and you don't really have to worry about that anymore. And then you deal with something else that's horrible, until you find someone who's an expert in that into that kind of stuff. So there's a lot of a lot of the instability of what you feel like when you first are doing this kind of kind of start taking away as you kind of build up that group of people, you know, who are supporting you, and helping you with your vision as well. So you know, I don't code as much as I do, because, like, it's my business to be the front end of the company and to talk about and you'll learn how to, to listen to customers. And so, you know, I don't spend as much time coding and my co founder, who's the CTO does all that kind of stuff. And so I don't have to make as many decisions there and so therefore that is a lot less stressful for me. He worries about you know, databases going down and all that kind of stuff like that. So yeah.
Nick Jikomes 1:04:47
I don't want to take too much more of your time. Dan, this has been fascinating. We talked about like your the science you used to do as well as this new startup which is really interesting. Is there anything else you want to say to people or or final thoughts you want to leave them with in terms of what you're doing and, uh, Rita or anything else that we talked about?
Daniel Brady 1:05:03
Yeah, I mean, I think I think it kind of ties a lot of what we're talking about together. And what we do at arena, but more globally, especially with what's going on, especially with all the hype is that like, you know, we both have science, strong science backgrounds. And what really makes a difference, when you want to use science or data to do something is that the information that you have is really good. That's what fundamentally matters. That's, you know, there everyone used to talk about a few years ago, big data, it's good data. Really, yes, people can make really, really fundamental discoveries about the universe, like force equals mass times acceleration, where like, not a ton of data, but there's like, had really, really great data. And they're really, really smart. And they're able to action on that. And I think we're like, increasingly, in this era, where there's a lot of audit information, there's a lot of lot of different tools, and machine learning and generative AI and all these sorts of things that can work on, and maybe transform the ways in which we interact with all sorts of things in our life. But they're all under the foot. Work with data. And if that fundamental data is not high quality, then none of it matters. And as you said, it's garbage in garbage out. So like, if you want to make a chatbot, if you want to do something else, if you want to make a decision on on who to make as a customer or you're trying to say like Who should we help, like all sorts of different reasons should we do this drug or not? What matters fundamentally, is that your data is high quality. And that's something that we do as a company as a standalone thing, but like, is a fundamental throughline of all the work that you both of us have done our entire lives. And so as, as you know, everyone goes out and listens to and reads about all the different exciting things that are happening in the machine learning space and in artificial intelligence, all that kinda stuff. Just remember what's actually feeding what is the fuel underneath that hood? And and to make sure that like before you believe any of the claims about what's going on that that's like a central component of it. And to be like very wary. If if, you know, something's like, oh, we spent up this thing. And we don't care about the data that it worked on. We don't care about the privacy or security implications, like all of that is like incredibly important.
Nick Jikomes 1:07:15
All right. Well, Dr. Daniel Brady, thank you for your time. Thank you so much.
Comments